Server monitoring, graphing, etc [somewhat for damien]

Damien and I have had a couple of short conversations about monitoring,
and nagios, and cacti. I've been looking into other monitoring tools as
of late, since Nagios, while nice for doing some things, isn't quite
giving me what I need at work, and snmp and mrtg just aren't jiving very
well. (Creating graphs that are based on non-interface data, via snmp,
is a pain.)

A coworker found a really neat little graphing/monitoring tool called
'collectd', which gathers data every 10 seconds (you can change it if
you edit collectd.c, but as of now, it's not supposed to be something
you change at all), and saves it to an rrd file. You run a contributed
perl script that generates some nice graphs.

It has quite a bit of plugins, and while generating graphs is rather
cpu-intensive, it's not half bad if you leave your server to generate
the graphs every half hour or so. Furthermore, it is worth noting that
if you are thinking of taking the RRDtool files, copying them to another
machine, and then making the graphs there, it won't work - at least, it
didn't for me. My 686 laptop wouldn't create graphs from amd64 rrd files.

If you wanna see more about collectd, you can read about it here:
http://collectd.org/

If you want to see some close-to-real-time graphs, you can go here:

https://ak.glaven.org/atropos.html # Laptop 'server'
https://ak.glaven.org/squeezy.html # Workstation

I plan on running it a while to see how things go - but so far, I'm
impressed. I can't give a very good example of how much it affects
performance since the laptop is slow and constantly swapping (450 MHz,
128MB RAM), and my workstation has been very busy doing things like
encoding video files all day. (Just check the graphs ;))

Adam

---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
Received on Fri May 12 16:05:22 2006